A Short Course on

Synchronous Machines and

Synchronous Condensers
G. Heydt S. Kalsi E. Kyriakides
Arizona State University
American Superconductor
© 2003 G. Heydt, S. Kalsi and E. Kyriakides
 Session Time Topics Instructor
Introductions 8:30 – 8:40 Bradshaw
1 Fundamentals of 8:40 – 9:50 • Energy conversion Heydt
synchronous • Synchronous machine
machines construction
• Energy transfer in a synchronous
machine
• Motor and generator action
• Phasor diagram for synchronous
machines
• Losses
• Superconducting designs
• Power factor and torque angle
• Example of calculations
• Transients and damper windings
• Saturation and the magnetization
curve
BREAK 9:50 – 10:00
2 Synchronous 10:00 – • What is a synchronous Kalsi
condensers 10:30 condenser?
• Applications of synchronous
condensers
• Analysis
3 Superconducting 10:30 – • Superconductivity Kalsi
synchronous 12:00 • The superconducting
condensers synchronous condenser (SSC)
• Performance benefits of SSC in a
grid
LUNCH 12:00 – 1:30
4 Synchronous 1:30 – 2:30 • Park’s transformation Heydt
machine models • Transient and subtransient
reactances, formulas for
calculation
• Machine transients
5 State estimation 2:30 – 3:30 • Basics of state estimation Kyriakides
applied to • application to synchronous
synchronous generators
generators • demonstration of software to
identify synchronous generator
parameters
 BREAK 3:30 – 3:40
6 Machine 3:40 – 4:30 • DFRs Heydt,
instrumentation • Calculation of torque angle Kyriakides,
• Usual machine instrumentation and Kalsi
Question and 4:30 – 5:00 All
answer session participants
 SESSION 1

Fundamentals of
synchronous machines
 Synchronous Machines
• Example of a rotating electric machine
• DC field winding on the rotor, AC armature
winding on the stator
• May function as a generator (MECHANICAL
Æ ELECTRICAL) or a motor (ELECTRICAL
Æ MECHANICAL)
• Origin of name: syn = equal, chronos =
time
 Synchronous Machines

ROTATION

• FIELD WINDING

• ARMATURE WINDING
 Synchronous Machines
The concept of air gap flux

STATOR

ROTOR
 Synchronous Machines
• The inductance of the stator winding
depends on the rotor position
• Energy is stored in the inductance
• As the rotor moves, there is a change in the
energy stored
• Either energy is extracted from the magnetic
field (and becomes mechanical energy – that
is, its is a motor)
• Or energy is stored in the magnetic field and
eventually flows into the electrical circuit that
powers the stator – this is a generator
 Synchronous Machines
The basic relationships are

POWER = ( TORQUE ) (SPEED)

ENERGY = (1/2) ( L I 2 )

POWER = d(ENERGY) / d(TIME)

 Synchronous Machines
Consider the case that the rotor (field) is energized
by DC and the stator is energized by AC of
frequency f hertz.

There will be average torque produced only when

the machine rotates at the same speed as the
rotating magnetic field produced by the stator.

RPM = ( 120 f ) / (Poles)

Example: f = 60 Hz, two poles, RPM = 3600 rev/min

 Synchronous Machines
d

The axis of the field winding

in the direction of the DC
field is called the rotor direct
axis or the d-axis. 90
degrees later than the d-axis
is the quadrature axis (q-axis). ROTATION

q
The basic expression for the voltage in the
stator (armature) is
v = r i + dλ/dt
Where v is the stator voltage, r is the stator
resistance, and λ is the flux linkage to the
field produced by the field winding
 Synchronous Machines

Basic AC power flow

jx
SEND RECEIVE

Vsend

Vreceive
 Synchronous Machines

Vinternal

Vterminal

The internal voltage,

often labeled E, is
GENERATOR ACTION –
produced by the field POWER FLOWS FROM
interacting with the MACHINE TO EXTERNAL
stator winding, and
this is the open circuit
CIRCUIT, E LEADS Vt
voltage
 Synchronous Machines
Vterminal
Vinternal

The internal voltage,

often labeled E, is
MOTOR ACTION – POWER
produced by the field FLOWS FROM EXTERNAL
interacting with the CIRCUIT INTO THE
stator winding, and
this is the open circuit
MACHINE, E LAGS Vt
voltage
 Synchronous Machines
Vinternal = E
TORQUE ANGLE
Vterminal = Vt
Vterminal = Vt TORQUE ANGLE
Vinternal = E
GENERATOR MOTOR

The internal voltage, often labeled E, is

produced by the field interacting with the
stator winding, and this is the open circuit
voltage
 Synchronous Machines

Active power will flow when there is a phase

difference between Vsend and Vreceive.
This is because when there is a phase
difference, there will be a voltage difference
across the reactance jx, and therefore there
will be a current flowing in jx. After some
arithmetic
Psent = [|Vsend|] [|Vreceive|] sin(torque angle) / x
 Synchronous Machines

Example
A synchronous generator stator
reactance is 190 ohms, and the internal
voltage (open circuit) generated is 35
kV line to line. The machine is
connected to a three phase bus whose
voltage magnitude is 35 kV line-line.
Find the maximum possible output
power of this synchronous generator
 Synchronous Machines
Example
Work on a per phase basis
35 kV line-line = 20.2 kV l-n
Max P occurs when torque angle is 90
degrees
P = (20.2K)(20.2K)(sin(90))/190
= 2.1 MW per phase = 6.3 MW three phase
Vsend

Vreceive
 Synchronous Machines

Example
If the phase angle is limited to 45 degrees, find
the generator power output
Vsend
TORQUE ANGLE

Vreceive
 Synchronous Machines

Example
P = 6.3 sin(45) = 4.6 MW

Vsend
TORQUE ANGLE

Vreceive
 Synchronous Machines
Losses
Rotor: resistance; iron parts moving in a magnetic
field causing currents to be generated in the rotor
body; resistance of connections to the rotor (slip
rings)
Stator: resistance; magnetic losses (e.g., hysteresis)
Mechanical: windage; friction at bearings, friction at
slip rings
Stray load losses: due to nonuniform current
distribution

EFFICIENCY = OUTPUT / INPUT

= 1 – (LOSSES) / INPUT
 Synchronous Machines
Losses
Generally, larger machines have the higher
efficiencies because some losses do not increase
with machine size. For example, many generators
in the 5 MW class and above have efficiencies
greater than 97%

But 3% of 5 MW is still 150 kW – and for large units

– e.g. 600 MW, 3% of 600 MW is 18 MW!

• Cooling
• Damping
 Power factor
Power factor is the cosine between
voltage and current in a sinusoidal
AC circuit.
Vsend = E

Voltage drop in reactance

Vreceive = Vt

GENERATOR Current in the circuit

NOTATION
 Power factor

Vsend

Voltage drop in reactance

Vreceive

Current in the circuit

Angle between
sending volts and
GENERATOR current
NOTATION
 Power factor

Vsend = E

Voltage drop in reactance

Vreceive = Vt

Current in the circuit

GENERATOR Angle between
NOTATION receiving volts and
COSINE OF THIS ANGLE IS current
THE MACHINE POWER
FACTOR AT THE TERMINALS
 Power factor
Current in the circuit
Angle between
receiving volts
Vreceive = Vt and current
Voltage drop in reactance
Vsend = E
COSINE OF THIS ANGLE IS
THE MACHINE POWER
FACTOR AT THE TERMINALS

MOTOR
NOTATION
 Power factor

Note that the power factor

angle is controllable by the
generated voltage E and hence
by the DC field excitation.
 Basic expressions

MOTOR Vt = E + jIax

GENERATOR Vt = E - jIax
 Power factor

Consider now a machine that:

1. Is operated at successively smaller
and smaller torque angle
2. Greater and greater field excitation
 Successively smaller and
smaller torque angle
The machine torque angle is made smaller and
smaller by reducing the electrical load (P)

Vt = E + jIax Current in the circuit

Vt

E
MOTOR
NOTATION Voltage drop in reactance
 Successively smaller and
smaller torque angle
The machine torque angle is made smaller and
smaller by reducing the electrical load (P)

Vt = E + jIax
Vt Current in the circuit

E
MOTOR
NOTATION Voltage drop in reactance
 Successively smaller and
smaller torque angle
The machine torque angle is made smaller and
smaller by reducing the electrical load (P)

Vt = E + jIax

Current in the circuit Vt

E
MOTOR
NOTATION Voltage drop in reactance
 Successively smaller and
smaller torque angle
The machine torque angle is made smaller and
smaller by reducing the electrical load (P)

Vt = E + jIax

Current in the circuit Vt

E
MOTOR
NOTATION Voltage drop in reactance
 Successively greater field
excitation
Increasing the field excitation causes E to
increase
Vt = E + jIax

Current
in the Vt
circuit

MOTOR
E
NOTATION Voltage drop in reactance
 Successively greater field
excitation
Increasing the field excitation causes E to
increase
Vt = E + jIax

Current
in the Vt
circuit

MOTOR
E
NOTATION Voltage drop in reactance
 Successively greater field
excitation
Increasing the field excitation causes E to
increase
Vt = E + jIax

Current
in the Vt
circuit

MOTOR
E
NOTATION Voltage drop in reactance
The foregoing indicates that as the machine
(1) approaches zero power operation – the
borderline between generator and motor
operation, the active power to/from the
machine goes to zero and (2) as the machine
becomes overexcited, the power factor
becomes cos(90) = 0.

As the field excitation increases, |E|

increases, and the machine current becomes
higher – but the power factor is still zero.
And I leads Vt. In theory, there is no active
power transferred, but a high and
controllable level of Q.

This mode of operation is called a

synchronous condenser
Synchronous condenser
operation
Ia
Q = | Vt |2 | I a |2 − P 2
= | Vt || I a |

V jIax
E
 Synchronous condenser
operation
Nearly zero active power flow,
nearly zero power factor, nearly
perpendicular Ia and Vt, current
leads terminal voltage acting as
a motor, it acts as a capacitor
Ia
jIax
Vt
Power factor correction,
reactive power support,
voltage support,
reactive power can be
varied by varying
excitation, low loss, no
‘resonance problems’ of
conventional fixed
capacitors, potentially a
large source of reactive
power
E
 Examples
A synchronous generator is rated 100 MVA.
The machine is intended to be operated at
rated power at torque angle = 37 degrees.
The armature resistance is 0.1%, and the
reactance is 85%. The terminal voltage is
rated 34.5 kV. Find the machine internal
percent excitation and terminal pf when the
machine operates at 100 MW. Estimate the
armature I2R losses.
Examples

| Vt || E | sin(δ )
P=
x
(1)(| E | sin(37 o )
=
0.85
= 1.41
 Examples
E = 1.41 /37o
Ia
Vt = 1.00 /0o

1.00∠0o + jI a (0.85) = 1.42∠37 o

0.85 | I a | ∠ϕ + 90o = 1.42∠37 − 1.00∠0o
| I a |= 1.02 φ = −8.9o
POWER FACTOR
cos(−8.9o ) = 98.8% LAGGING
 Examples

A six pole synchronous generator

operates at 60 Hz. Find the speed of
operation
 Examples
RPM = ( 120 f ) / (Poles)

RPM = 120*60 / 6

= 1200
 Examples

A 40 MVAr synchronous condenser

operates on a 34.5 kV bus. The
synchronous reactance is 150%.
Estimate the field excitation to obtain a
30 to 40 MVAr range of reactive power.
 Examples
0.75 ≤ Q ≤ 1.0
0.75 ≤} | I a |≤ 1.0
Vt = E + jI a x
1∠0 =| E f | ∠0 + jI a (1.5)
[1.0− | E f | ∠0 = 1.5 | I a | ∠(90 + 90 )
o o

At | I a |= 0.75
1.0− | E f |= −1.5 | 0.75 |
| E f |= 2.125
 Examples
Vt = E + jI a x
1∠0 =| E f | ∠0 + jI a (1.5)
[1.0− | E f | ∠0 = 1.5 | I a | ∠(90 + 90 )
o o

At | I a |= 1.00
1.0− | E f |= −1.5 | 1.00 |
| E f |= 2.50

Therefore the field excitation should be

between 213% and 250 %
 SESSION 4

Synchronous machine
models
• Saturation and the
magnetization curve
• Park’s transformation
• Transient and subtransient
reactances, formulas for
calculation
• Machine transients
Saturation and the magnetization
curve

SHORT CIRCUIT ARMATURE CURRENT

OPEN CIRCUIT TERMINAL VOLTAGE

RATED Vt OCC SCC

NE
RATED Ia
LI
AP
G

c
R
AI

f’ f’’
FIELD EXCITATION
 SHORT CIRCUIT ARMATURE CURRENT
OPEN CIRCUIT TERMINAL VOLTAGE

RATED Vt OCC SCC

NE
RATED Ia
LI
AP
G

c
R
AI

f’ f’’
FIELD EXCITATION

SYNCHRONOUS REACTANCE =
SLOPE OF AIR GAP LINE
SHORT CIRCUIT RATRIO = Of’/Of’’
Saturation and the magnetization
curve
• Saturation occurs because of the alignment
of magnetic domains. When most of the
domains align, the material saturates and no
little further magnetization can occur
• Saturation is mainly a property of iron -- it
does not manifest itself over a practical
range of fluxes in air, plastic, or other non-
ferrous materials
• The effect of saturation is to lower the
synchronous reactance (to a ‘saturated
value’)
Saturation and the magnetization
curve
• Saturation may limit the performance of
machines because of high air gap line
voltage drop
• Saturation is often accompanied by
hysteresis which results in losses in
AC machines
• Saturation is not present in
superconducting machines
 Transients and the dq
transformation
rF
ia

iF LF a
vF

ra
rD
L aa
L bb ib va
iD LD rb
v D =0
b
rQ
rn Lc c
vb
iQ LQ Ln rc
v Q =0 vn ic
c
rG
vc

iG LG n
v G =0
in
 Transients and the dq
rF
transformation
ia

iF LF a
vF

ra
rD
L aa
L bb ib va
iD LD rb
v D =0
b
rQ
rn Lc c
vb
iQ LQ Ln rc
v Q =0 vn ic
c
rG
vc

iG LG n
v G =0
in


v = − ri − λ
Transients and the dq
transformation
d-axis THE VOLTAGE EQUATION AS
THE BASIC IDEA IS TO WRITE

IF THERE WERE ONLY A d-

AXIS, AND AGAIN AS IF
THERE WERE ONLY A q-AXIS

ROTATION
q-axis


v = − ri − λ
 Transients and the dq
transformation

v = − ri − λ
 
λ 
 va   ra 0 0 0 0 0 0   ia  a
 v  0     
rb 0 0 0 0 0 ib  λb 
 b    
 vc  0 0 rc 0 0 0 0   i c   λc 
− v  = −  0     
0 0 rF 0 0 0 i F −  λF  +
 F   
− v D  0 0 0 0 rD 0 0  i D   λ D 
− v  0     
0 0 0 0 rG 0 iG  λ 
 G    G
 − v Q   0 0 0 0 0 0 rQ  i Q   λ 
 Q
 Transients and the dq
transformation
 1 1 1 
 
 2 2
π
2
π
 PARK’S TRANSFORMATION
P= cos θ cos(θ − ) cos(θ +
3  3 
2 2 2 )
3
 sin θ sin(θ − 2π 3 ) sin(θ + 2π 3 ) 
 

θ = ωR t + δ + π 2 BY APPLYING PARK’S
TRANSFORMATION, THE TIME
VARYING INDUCTANCES
BECOME CONSTANTS
 Transients and the dq
transformation
 v0  r + 3rn 0 0 0 0 0 0   i0 
 v   0 r ω( LAQ + A q ) 0 0 ωLAQ ωLAQ   id 
 d
   
 vq   0 − ω( LAD + A d ) r − ωLAD − ωLAD 0 0   iq 
− v  = −  0 0 0 rF 0 0 0   iF 
 F
   
 − vD   0 0 0 0 rD 0 0  iD 
− v   0 0 0 0 0 rG 0  iG 
 G
   
 − vQ   0 0 0 0 0 0 rQ  iQ 

 L0 + 3Ln 0 0 0 0 0 0   i0 
 0 LAD + A d 0 LAD LAD 0 0   i 
  d 
 0 0 LAQ + A q 0 0 LAQ LAQ   iq 
−
1 
⋅ 0 LAD 0 LAD + A F LAD 0 0  i 
ωB   F 
 0 LAD 0 LAD LAD + A D 0 0  iD 
 0 0 LAQ 0 0 LAQ + A G LAQ  iG 
  
 0 0 LAQ 0 0 LAQ LAQ + A Q  iQ 
 Machine reactances
rF LF ra La

iF rD id
vF iD
+ LAD
vd
LD
+

ωψq
d-axis equivalent circuit
Machine reactances
rG LG ra La

iG rQ iq
iQ
LAQ
vq
LQ
+

ωψd
q-axis equivalent circuit
 Machine reactances
• These equivalent circuit parameters are
traditionally obtained by a combination of
manufacturers’ design specifications and
actual tests
• IEEE has a series of standardized tests for
large generators that yield several time
constants and equivalent circuit inductances
• Aging and saturation are not well accounted
• Change in operating point is not well
accounted
 Machine transient and subtransient
reactances
Subtransient direct axis LD L2AD + LF L2AD − 2L3AD
inductance L "
d Ld −
LF LD − L2AD
Transient direct axis L2AD
inductance L '
d Ld −
LF
Subtransient open
circuit time constant in τ "
do
LD LF − L2AD
the direct axis ω B rD LF
Transient open circuit
time constant in the τ '
do
LF
direct axis ω B rF
Subtransient short
circuit time constant in τ "
d
L"d "
τ do
the direct axis Ld
'

Transient short circuit

time constant in the τ '
d
L'd '
τ do
direct axis Ld
Ea’ jiqxq SYNCHRONOUS
GENERATOR
jidxd PHASOR DIAGRAM

iq iara
ia
Vt

id
Ea’ jiqxq SYNCHRONOUS
GENERATOR
jidxd PHASOR DIAGRAM

iq iara
POWER ia
FACTOR
Vt ANGLE

TORQUE
ANGLE

id
Ea’ jiqxq SYNCHRONOUS
jiqxq ji x GENERATOR
PHASOR DIAGRAM
jiaxq d d
idxq
iq iara
POWER ia
FACTOR
Vt ANGLE

TORQUE
ANGLE

id
Machine transient and subtransient
reactances

The usual procedure is that IEEE

standardized tests are used to obtain
inductances and time constants. Then
using the formulas, circuit inductances
and resistances can be solved.

TIME CONSTANTS EQUIVALENT

TESTS INDUCTANCES CIRCUIT
PARAMETERS
 Transient calculations
• Transients in dynamic systems are calculated as solutions
of differential equations
• The usual solution approach is a numerical solution of
(dX/dt) = AX + bu
• Most numerical solutions relate to the approximation of
dX/dt as (delta X)/(delta t)
• Solutions are iterative in the sense that the given initial
condition is used to obtain X at time t = h; then X(h) is
used to obtain X(2h), etc.
• Popular solution methods include Matlab toolboxes, EMTP,
ETMSP, PSpice
• The computer solutions could be used to compare with
actual field measurements. And if there are discrepancies,
the computer model could be updated to obtain better
agreement – and hence a more accurate model.
 SESSION 5

State estimation applied to

synchronous generators
Session topics:

• Basics of state estimation

• Application to synchronous generators
• Demonstration of software to identify
synchronous generator parameters
 BASICS OF STATE ESTIMATION
+ R1=5Ω
V s=10V -
It is desired to measure
V2 V the voltage across R2
R2=5Ω

Assume we have two voltmeters: A and B Va = 5.1 V

Measure the voltage across R2 with both Vb = 4.7 V
voltmeters

Since the two measurements do not agree but are close to

each other, average the result to estimate V2
Va + Vb 5.1 + 4.7
V2 = = = 4. 9 V
2 2
 BASICS OF STATE ESTIMATION
Now assume that we have a third voltmeter C
Let the measurement from C be Vc = 15 V
Clearly this measurement is not reliable
Simple approach: disregard Vc and estimate V2 from Va and Vb
Another approach: Use weighted state estimation
This means, assign appropriate weights to each of the three
measurements according to the confidence that the user has
to each instrument.
For example, give the following weights:
• if B is the best instrument give it a weight of 20
• give a weight of 18 to A
• give a weight of 1 to C since it is not reliable
5.1× 18 + 4.7 × 20 + 15 × 1
⇒ V2 = = 5.15
39
 BASICS OF STATE ESTIMATION
Definition: State estimation is the process of assigning a
value to an unknown system state variable, using
measurements from the system under study. Knowledge
of the system configuration and of the accuracy of the
measuring instruments is used in this process.

Measurements

Estimator
System

Estimated
z x̂ states
H
 EXAMPLE 1
Assume that it is desired to estimate two states (variables)
Three measurements are obtained, which form the following
equations
x1 + x2 = 3.1 1 1   3.1 
2 x1 − x2 = 0.2 In matrix form: 2 − 1  x1  =  0.2 
   x2   
x1 − 3 x2 = −4.8 1 − 3 − 4.8

Process matrix 2 states 3 measurements

3x2 2x1 vector 3x1 vector

The matrix equation is of the form Hx = z

 EXAMPLE 1
1 1   3.1  Number of measurements: n=3
2 − 1  x1  =  0.2  Number of states: m=2
   x2    Since n>m, the system is overdetermined
1 − 3 − 4.8 Hence there is no unique solution
The solution is not unique since in general it is not possible to
satisfy all the equations exactly for any choice of the
unknowns.
A solution should be selected such that the error in satisfying
each equation is minimum.
This error is called the residual of the solution and can be
computed by,
r = z − Hxˆ
x̂ : the vector of the estimated parameters
The residual will be calculated later
 EXAMPLE 1
There are many ways to minimize the residual r
One of the most popular is the least squares method, which
in effect minimizes the length (Euclidean norm) of the
residual r.
This method results in a simple formula to calculate the
estimated parameters
Given the system is of the form Hx=z,the vector of the
estimated parameters is given by,

ˆx = ( H T H ) −1 H T z = H + z

H+ is called the pseudoinverse of H

 H x z EXAMPLE 1

1 1   3.1  Substitute H and z in xˆ = ( H T H ) −1 H T z

2 − 1  x1  =  0.2 
   x2    and solve for the unknown states
1 − 3 − 4.8
−1
 1 1    3.1 
 1 2 1     1 2 1   
xˆ =  2 − 1 0. 2
 1 − 1 − 3    1 − 1 − 3  
 1 − 3  − 4.8
−1
 6 − 4  − 1.3 0.22 0.08 − 1.3
⇒ xˆ =     =  17.3 
 − 4 11   17 . 3   0 . 08 0. 12  
1.098
⇒ xˆ =  
1 .972 
 EXAMPLE 1
To see how much error we have in the estimated parameters,
we need to calculate the residual in a least squares sense

T T
J = r r = ( z − Hxˆ ) ( z − Hxˆ )
1 1   3.1   − 0.03 
  1.098 
r = Hx − z = 2 − 1 
ˆ  − 0.2  =  0.224 
  1.972    
1 − 3 − 4.8 − 0.018

 − 0.03 
⇒ J = [− 0.03 0.224 − 0.018] 0.224  = 0.0514
 
− 0.018
 WHY ARE ESTIMATORS NEEDED?
In power systems the state variables are
typically the voltage magnitudes and the relative
phase angles at the nodes of the system.

The available measurements may be voltage

magnitudes, current, real power, or reactive
power.

The estimator uses these noisy, imperfect

measurements to produce a best estimate for
the desired states.
 WHY ARE ESTIMATORS NEEDED?
It is not economical to have measurement devices at
every node of the system

The measurement devices are subject to errors

If errors are small, If errors are large,

these errors may go the output would be useless
undetected

There are periods when the communication

channels do not operate. Therefore, the system
operator would not have any information about
some part of the network.
HOW DOES THE ESTIMATOR HELP?

An estimator may:
• reduce the amount of noise in the measurements
• detect and smooth out small errors in readings
• detect and reject measurements with gross errors
• fill in missing measurements
• estimate states that otherwise are difficult to
measure
 EXAMPLE 2
V1 R1
Assume we have a network
V2 R2
configuration as in the figure on the left.
V3
Assume that measurements are
available for V1, V2, and V3. Find a
R3
relationship for V3 that has the following
form: V3 = aV1 + bV2 + c
Available measurements This is clearly an estimation
V1 V2 V3 problem with three unknowns (a,
7.1 0 3.1 b, c), and four measurements.
8.3 3.2 2.3 Therefore it is an overdetermined
10.4 5.1 1.4
0 9.1 4.0
estimation problem.

Now, it is necessary to express the estimation problem

mathematically
 EXAMPLE 2
Substitute the measurements obtained in the desired model
V3 = aV1 + bV2 + c
V1 V2 V3
7.1 0 3.1
3.1 = 7.1a + 0b + c 8.3 3.2 2.3
2.3 = 8.3a + 3.2b + c 10.4 5.1 1.4
1.4 = 10.4a + 5.1b + c 0 9.1 4.0
4.0 = 0a + 9.1b + c
In matrix form, As in example 1, we can solve this
 7.1 0 1  3.1 matrix equation by taking the
 8.3 3.2 1 a   2.3 pseudoinverse of the H matrix
  b  =    7.1 0 1
+
 3.1
10.4 5.1 1   1.4  a  
ˆ  2.3  − 0.3028
 0 9.1 1  c  4.0   1
  = − 0.1374
8.3 3.2
    ⇒ b =ˆ  
  10.4 5.1 1 1.4   
 cˆ   4.0  5.2505 
 0 9.1 1  
x z
H
 EXAMPLE 3
Let’s work out another example:
Estimate the relative phase angles at the buses of the figure
below

40 MW
BUS 1 BUS 2
M12 20 MW M32

60 MW
50 MW
90 MW

M13
30 MW BUS 3
Given:
X12 = 0.2 p.u.
X13 = 0.4 p.u.
X23 = 0.1 p.u.
System base: 100 MVA
 EXAMPLE 3
SOLUTION

40 MW
BUS 1 BUS 2
M12 20 MW M32 Let bus 1 be the reference bus

60 MW
50 MW
90 MW
ϑ1 = 0

From the measurements:

M13
30 MW BUS 3 M12 = 20 MW = 0.2 p.u.
M13 = 30 MW = 0.3 p.u.
The line flows are given by, M32 = -60 MW = -0.6 p.u.
1
f ab = (ϑa − ϑb ) = M ab
X ab V1 ∠δ1 V2 ∠δ2
The above formula can be shown X
considering a simple two bus arrangement P
V1 V2 sin(δ1 − δ2 ) Since V1 and V2 are approximately 1 p.u., and the
P≈
X angle δ1 − δ2 is small, P can be obtained as,
δ −δ
P≈ 1 2
X
 EXAMPLE 3
Hence, f12 =
1
(ϑ1 − ϑ2 ) =
1
(0 − ϑ2 ) = −5ϑ2 = 0.2
X 12 0.2
1 1
f13 = (ϑ1 − ϑ3 ) = (0 − ϑ3 ) = −2.5ϑ3 = 0.3
X 13 0.4
1 1
f 32 = (ϑ3 − ϑ2 ) = (ϑ3 − ϑ2 ) = 10ϑ3 − 10ϑ2 = −0.6
X 32 0. 1

 −5 0   0 .2 
ϑ
In matrix form,  0 − 2.5  2  =  0.3 
ϑ
− 10 10   3  − 0.6

This is again of the form Hx = z, and is solved as in example 1:

−1
  −5 0   0 .2 
 − 5 0 − 10    − 5 0 − 10  
ϑˆ =    0 − 2 . 5   0 . 3
 0 − 2.5 10    0 − 2.5 10   
 − 10 10   − 0.6


ϑˆ2  − 0.0438
⇒ =  rad
ϑ3  
ˆ − 0 .1048 
APPLICATION OF STATE ESTIMATION
TO SYNCHRONOUS GENERATORS
• Need to know the operating parameters of generators to
→ perform studies
→ study behavior of the system at various operating levels
→ perform postmortem analysis
• Meet requirements for machine testing (e.g. NERC)
• To reestablish machine parameters after a repair
• Fault identification / signature analysis
• Incipient event identification
APPLICATION OF STATE ESTIMATION
TO SYNCHRONOUS GENERATORS
Problems:
Generator parameters change with operating point, aging
Cannot measure parameters while generator is committed
Cannot afford to decommit unit in order to measure its
parameters
Solution:
Use available terminal measurements, knowledge of the
model of the generator, and state estimation, to approximate
the required parameters

To do that, it is necessary to develop a model for the

synchronous generator
 SYNCHRONOUS GENERATOR
REPRESENTATION
a axis

d axis ϑ q axis
ib ic
direction of
rotation iQ
iG iD

ia iF ia
iF iG
iD iQ

b axis ic ib c axis
SYNCHRONOUS GENERATOR MODEL
rF
ia
vF iF LF a

rD ra
Laa ib
iD Lbb r va
vD=0 LD
b
b
rQ
rn Lcc
vb
vQ=0
iQ LQ Ln vn rc
ic
c
rG
vc
iG LG n
vG=0
in

Schematic diagram of a synchronous generator

 DEVELOPMENT OF SYNCHRONOUS
GENERATOR MODEL
 va   ra 0 0 0 0 0 0   ia   λa 
 v  0  
rb 0 0 0 0 0   ib   λb 
 b
   
0   ic   λc 
v = − ri − λ − vn
 vc  0 0 rc 0 0 0
− v  = −  0   v 
⇒ 0 0 rF 0 0 0  iF  −  λF  +  n 
 F   
0  iD   λD   
0
−
 Dv 0 0 0 0 rD 0
 − vG  0  
0 0 0 0 rG 0  iG   λG 
    
− v
 Q   0 0 0 0 0 0 rQ  iQ   λ 
 Q

 λa   Laa Lab Lac LaF LaD LaG LaQ   ia   λ0   L0 0 0 0 0 0 0   i0 

λ   L Lbb Lbc LbF LbD LbG LbQ   ib  λ   0 Ld 0 kM F kM D 0 0   id 
 b   ba    d   
 λc   Lca Lcb Lcc LcF LcD LcG LcQ   ic   λq   0 0 Lq 0 0 kM G kM Q   iq 
λ  =  L LFQ  iF  λ  =  0  i F 
 F   Fa
LFb LFc LFF LFD LFG
  ⇒  F 
0 kM F 0 LF M X 0
 
 λD   LDa LDb LDc LDF LDD LDG LDQ  iD   λD   0 kM D 0 M X LD 0 0  iD 
 λG   LGa LGb LGc LGF LGD LGG LGQ  iG   λG   0 0 kM G 0 0 LG M Y  iG 
         
 λQ   LQa LQb LQc LQF LQD LQG LQQ  iQ   λQ   0 0 kM Q 0 0 M Y LQ  iQ 
 DEVELOPMENT OF SYNCHRONOUS
GENERATOR MODEL
λ0 dq = Pλabc i0 dq = Piabc v0 dq = Pvabc

1 1 1 
 2 2 2 
P = 2  cos ϑ cos(ϑ − 2π ) cos(ϑ + 2π )
3 3 3 
 sin ϑ sin(ϑ − 2π ) sin(ϑ + 2π ) 
 3 3 

Resulting model:
 v0 dq 3 x1   i0 dq 3 x1   i0 dq 3 x1 
v  = −[R7 x 7 ]i  − [L7 x 7 ]i 
 FDGQ 4 x1   FDGQ 4 x1   FDGQ 4 x1 
SYNCHRONOUS GENERATOR MODEL
 v0  r + 3rn 0 0 0 0 0 0   i0 
 v   0 r ω( LAQ + A q ) 0 0 ωLAQ ωLAQ   id 
 d
   
 vq   0 − ω( LAD + A d ) r − ωLAD − ωLAD 0 0   iq 
− v  = − 0 0 0 rF 0 0 0  i F 
F
    
−
 Dv  0 0 0 0 rD 0 0  iD 
 − vG   0 0 0 0 0 rG 0  iG 
    
− v
 Q   0 0 0 0 0 0 rQ  iQ 
 L0 + 3Ln 0 0 0 0 0 0   i0 
 0 LAD + A d 0 LAD LAD 0 0   i 
  d 
 0 0 LAQ + A q 0 0 LAQ LAQ   iq 
−
1 
0 LAD 0 LAD + A F LAD 0 0  i 
ωB   F 
 0 LAD 0 LAD LAD + A D 0 0  iD 
 0 0 LAQ 0 0 LAQ + A G LAQ  iG 
  
 0 0 L AQ 0 0 L AQ L AQ + A  iQ 
Q
 MODEL DISCUSSION
After the development of the model it is necessary to
carefully examine the available information about the
system, find out what is known in the model, what is
unknown and needs to be calculated or assumed, and what
is desired to be estimated.
For the synchronous generator case,
Measured/Known Unknown
line-to-line terminal voltages damper currents
line currents current derivatives
field voltage (for an exciter with brushes)
field current (for an exciter with brushes)

Finally, some of the parameters need to be estimated through

state estimation, while the other parameters need to be
calculated from manufacturer’s data
STATE ESTIMATOR CONFIGURATION
EXAMPLE
Estimate LAD, LAQ, and rF
i (t + ∆t ) − i (t )
• Calculate current derivatives by using i(t ) ≈
∆t
• Rearrange system in the form Hx = z
 i0 
 0 0 0
i 
 1 ′ ′ ′ 
ω (id + i F + i D ) ω (i q + iG + iQ ) 0  L  r + 3rn 0 0 0 0 0 0  d 
 0 i 
 B   AD  r ωA q 0 0 0 0  q 
 − ω(id + iF + iD ) 1 L = −   i
(iq′ + iG′ + iQ′ ) 0   AQ   0 − ωA d r 0 0 0 0  F 
 ωB  r  i
 1  F   0
 0 0 0 0 0 0  D 
 (id′ + iF′ + iD′ ) 0 iF  i
 ωB   G
iQ 
 i0   V0 
 i   V 
 L0 + 3Ln 0 0 0 0 0 0  d   d 
  i   V 
1  0 Ad 0 0 0 0 0  q   q 
−  i − − V
ωB  0 0 Aq 0 0 0 0  F   F 
i  − V
 0
 0 0 AF 0 0 0  D   D 
i −V
 G  G
iQ   − VQ 
 DEMONSTRATION OF PROTOTYPE
APPLICATION FOR PARAMETER ESTIMATION

•Prototype application developed in Visual C++

•Portable, independent application
•Runs under Windows
•Purpose: Read measurements from DFR and use
manufacturer’s data to estimate generator
parameters
 SESSION 6

Machine instrumentation
Session topics:

•Digital Fault Recorders (DFRs)

• Calculation of torque angle
DIGITAL FAULT RECORDERS (DFRs)
A DFR is effectively a data acquisition system that is used to
monitor the performance of generation and transmission
equipment.
It is predominantly utilized to monitor system performance
during stressed conditions. For example, if a lightning
strikes a transmission line, the fault recognition by
protective relays and the fault clearance by circuit breakers
takes only about 50 to 83 ms.
This process is too fast for human intervention. Therefore,
the DFR saves a record of the desired signals (e.g. power
and current), and transmits this record to the central offices
over a modem, where a utility engineer can perform post-
event analysis to determine if the relays, circuit breakers and
other equipment functioned properly.
DIGITAL FAULT RECORDERS (DFRs)

The DFR
sends the
measured
signals to a
central pc
station
through a
modem
DIGITAL FAULT RECORDERS (DFRs)

Typical graphics window showing a snapshot of the

measured signals
 TYPICAL DFR SPECIFICATIONS

Data files are stored in COMTRADE IEEE format

The DFR can be configured to create transient records and
continuous records
Can be used during disturbances, abnormal conditions, and
normal conditions

Typical specifications:
Analog channels: 8, 16, 24, or 32
Digital channels: 16, 32, 48, or 64
Sample rate: 24-192 samples/min
Operating voltage: 48VDC, 125VDC, 250VDC, 120VAC
CALCULATION OF TORQUE ANGLE
The torque angle δ is defined as the angle between the
machine emf E and the terminal voltage V as shown in the
phasor diagram

δ V jxqIa
φ rIa

Ia
 CALCULATION OF TORQUE ANGLE
The torque angle can be calculated in
different ways depending on what information
is available
Two ways to calculate the torque angle will be
shown:
1. Using line to line voltages and line currents
(stator frame of reference)
2. Using voltages and currents in the rotor frame of
reference (0dq quantities)
 CALCULATION OF TORQUE ANGLE
IN THE STATOR REFERENCE FRAME
Known quantities:
Line to line voltages (vab, vbc, vca)
Line currents (ia, ib, ic)
Procedure:
1. Calculate phase voltages 2. Calculate three phase active
1
va = (vab − vca )
and reactive power
3 P = vabia − vbc ic
1
vb = (−vab + vbc ) Q = (vabic + vbc ia + vca ib ) 3
3
1
vc = (−vbc + vca )
3
3. Calculate the power factor
Q
φ = tan −1
P
 CALCULATION OF TORQUE ANGLE
IN THE STATOR REFERENCE FRAME
4. Calculate the voltage angle for each phase
For a balanced 3-phase system,
va = vm cos θ
vb = vm cos(θ − 120)
vc = vm cos(θ + 120)
For phase a, use phases b and c
1 3
vb = vm cos θ cos120 + vm sin θ sin 120 = − vm cos θ + vm sin θ
2 2
1 3
vc = vm cos θ cos120 − vm sin θ sin 120 = − vm cos θ − vm sin θ
2 2

It can be observed that, vb + vc = −vm cos θ ⇒

vb + vc
=−
1
cot θ
vb − vc = 3vm sin θ vb − vc 3
1 vb − vc
⇒ tan θ = −
3 vb + vc
 CALCULATION OF TORQUE ANGLE
IN THE STATOR REFERENCE FRAME
The angles for the other phases can be calculated in a
similar fashion. The angles for all phases are given by,
−1 − 1 (vb − vc )
θva = tan ( )
3 (vb + vc )
− 1 (vc − va )
θvb = tan −1 ( )
3 (vc + va )
− 1 (va − vb )
θvc = tan −1 ( )
3 (va + vb )

5. Find the angle of ia

Using the above procedure,
− 1 (ib − ic )
θia = tan −1 ( )
3 (ib + ic )
 CALCULATION OF TORQUE ANGLE
IN THE STATOR REFERENCE FRAME
6. Calculate the instantaneous line to neutral rms voltage for
phase a
vbc + vca
Vt rms = −
2 3 cosθ va

7. Calculate the machine generated emf

Et = Vt + (r + jxq ) I a = Et ∠δ

δ is the torque angle

 CALCULATION OF TORQUE ANGLE
IN THE ROTOR REFERENCE FRAME
Known quantities:
0dq voltages (v0, vd, vq)
0dq currents (i0, id, iq)

Procedure:
1. Calculate the active and reactive power
P = vd id + vq iq
Q = vq id − vd iq

2. Calculate the terminal voltage

Et = Vd + jVq = Et ∠γ
 CALCULATION OF TORQUE ANGLE
IN THE ROTOR REFERENCE FRAME
3. Calculate the terminal current
P2 + Q2
It =
Et

4. Calculate the power factor angle

P
−1
φ = cos ( )
Et I t
5. Calculate the torque angle
−1 ( xq I t cos φ − rI t sin φ)
δ = tan
( Et + rI t cos φ + xq I t sin φ)
 SESSION 7

Question and answer

session